Validating Interests for a Search and Feed Service

1. A system, comprising: a processor configured to: determine a curation score associated with an entity based at least in part on a plurality of tags applied to the entity by a plurality of different curator users, the entity corresponding to a potential user interest and the curation score indicating a first likelihood that a threshold number of a plurality of users will be interested in viewing web documents associated with the entity; determine a good interest probability value associated with the entity based at least in part on one or more feature data values corresponding to the entity, the one or more feature data values being derived from a data source comprising information regarding the entity, wherein the good interest probability value indicates a second likelihood that the threshold number of the plurality of users will be interested in viewing the web documents associated with the entity; and generate a content feed for a user of the plurality of users, the content feed including one or more web documents selected based in part on the curation score associated with the entity and the good interest probability value associated with the entity, wherein the entity is distinct from the one or more web documents; and a memory coupled with the processor, wherein the memory is configured to provide the processor with instructions.

2. The system of claim 1 , wherein the processor is further configured to: aggregate the plurality of tags to determine the curation score.

3. The system of claim 1 , wherein the data source comprises an information website.

4. The system of claim 3 , wherein the one or more feature data values includes an inverse document frequency (IDF) value associated with the entity.

5. The system of claim 3 , wherein the one or more feature data values includes a IDF value associated with one or more entities that resolve to the entity.

6. The system of claim 3 , wherein the one or more feature data values includes a link probability value associated with the entity.

7. The system of claim 3 , wherein the one or more feature data values includes a views over inlinks value associated with the entity.

8. The system of claim 3 , wherein the one or more feature data values includes a ratio of inlinks of a web document associated with the entity to outlinks of the web document associated with the entity.

9. The system of claim 3 , wherein the one or more feature data values includes a lowercase frequency value.

10. The system of claim 3 , wherein the one or more feature data values includes a log value of users subscribed to a page associated with the entity.

11. The system of claim 3 , wherein the good interest probability value is computed at least in part in applying the one or more feature data values corresponding to the entity to a machine learning model.

12. The system of claim 1 , wherein the processor is further configured to: determine whether the curation score associated with the entity is consistent with the good interest probability value associated with the entity; adjust the good interest probability value to be more consistent with the curation score based at least in part on a number of the plurality of different curator users, when the curation score is determined to be inconsistent with the good interest probability value; and generate the content feed based at least in part on the adjusted good interest probability value.

13. A method, comprising: determining a curation score associated with an entity based at least in part on curator user feedback, the curation score indicating a first likelihood that a threshold number of users will be interested in viewing content associated with the entity; determining a good interest probability value associated with the entity based at least in part on one or more statistical values associated with the entity, the one or more statistical values being independent of the curator user feedback, and the good interest probability value indicating a second likelihood that the threshold number of users will be interested in viewing content associated with the entity: and generating a content feed for a user that includes one or more web documents based in part on the curation score associated with the entity and the good interest probability value associated with the entity, wherein the entity is distinct from the one or more web documents.

14. The method of claim 13 , wherein the curation score is based on aggregated curator user feedback.

15. The method of claim 13 , wherein the one or more statistical values associated with the entity comprises one or more feature data values associated with the entity.

16. The method of claim 15 , wherein the good interest probability value is computed at least in part in applying the one or more feature data values associated with the entity to a machine learning model.

17. The method of claim 15 , wherein the one or more feature data values includes a link probability value associated with the entity.

18. The method of claim 15 , wherein the one or more feature data values includes a views over inlinks value associated with the entity.

19. The method of claim 13 , further comprising: determining whether the curation score associated with the entity is consistent with the good interest probability value associated with the entity; adjusting the good interest probability value based at least in part on a number of curator users that provided the curator user feedback when the curation score is inconsistent with the good interest probability value; and generating the content feed based at least in part on the adjusted good interest probability value.

20. A computer program product, the computer program product being embodied in a tangible non-transitory computer readable storage medium and comprising computer instructions for: determining a curation score associated with an entity based at least in part on aggregated curator user feedback the curation score indicating a first likelihood that a threshold number of users will be interested in viewing content associated with the entity; determining a good interest probability value associated with the entity based at least in part on applying computed values associated with the entity to a machine learning model, the good interest probability value indicating a second likelihood that the threshold number of users will be interested in viewing the content associated with the entity: and generating a content feed for a user that includes one or more web documents based in part on an evaluated consistency between the curation score associated with the entity and the good interest probability value associated with the entity, wherein the entity is distinct from the one or more web documents.